AI Feature Functions
gs_index_advise(text)
Description: Recommends an index for a single query statement.
Parameter: SQL statement string
Return type: record
For details, see section "DBMind: Autonomous Database O&M > AI Sub-functions of DBMind > Index-advisor: Index Recommendation > Single-Query Index Recommendation" in Feature Guide.
hypopg_create_index(text, [text])
Description: Creates a virtual index.
Parameter: (optional) character string of the statement for creating an index, level of the created virtual index
Return type: record
For details, see section "DBMind: Autonomous Database O&M > AI Sub-functions of DBMind > Index-advisor: Index Recommendation > Virtual Index" in Feature Guide.
hypopg_display_index([text])
Description: Displays information about all created virtual indexes.
Parameter: (optional) level of the virtual index to be displayed
Return type: record
For details, see section "DBMind: Autonomous Database O&M > AI Sub-functions of DBMind > Index-advisor: Index Recommendation > Virtual Index" in Feature Guide.
hypopg_drop_index(oid)
Description: Deletes a specified virtual index.
Parameter: OID of the index
Return type: Boolean
For details, see section "DBMind: Autonomous Database O&M > AI Sub-functions of DBMind > Index-advisor: Index Recommendation > Virtual Index" in Feature Guide.
hypopg_reset_index([text])
Description: Clears all virtual indexes.
Parameter: (optional) level of the virtual index to be cleared
Return type: none
For details, see section "DBMind: Autonomous Database O&M > AI Sub-functions of DBMind > Index-advisor: Index Recommendation > Virtual Index" in Feature Guide.
hypopg_estimate_size(oid)
Description: Estimates the space required for creating a specified index.
Parameter: OID of the index
Return type: int8
For details, see section "DBMind: Autonomous Database O&M > AI Sub-functions of DBMind > Index-advisor: Index Recommendation > Virtual Index" in Feature Guide.
check_engine_status(ip text, port text)
Description: Tests whether a predictor engine provides services on a specified IP address and port number.
Parameter: IP address and port number of the predictor engine.
Return type: text
This function is unavailable in the current version.
encode_plan_node(optname text, orientation text, strategy text, options text, dop int8, quals text, projection text)
Description: Encodes the plan operator information in the input parameters.
Parameter: plan operator information
Return type: text
This function is an internal function. You are advised not to use it directly.
model_train_opt(template text, model text)
Description: Trains a given query performance prediction model.
Parameters: template name and model name of the performance prediction model
Return type: startup_time_accuracy FLOAT8, total_time_accuracy FLOAT8, rows_accuracy FLOAT8, peak_memory_accuracy FLOAT8
This function is unavailable in the current version.
track_model_train_opt(ip text, port text)
Description: Returns the training log address of the predictor engine with the specified IP address and port number.
Parameter: IP address and port number of the predictor engine
Return type: text
This function is unavailable in the current version.
encode_feature_perf_hist(datname text)
Description: Encodes historical plan operators collected in the target database.
Parameter: database name
Return type: queryid bigint, plan_node_id int, parent_node_id int, left_child_id int, right_child_id int, encode text, startup_time bigint, total_time bigint, rows bigint, and peak_memory int
gather_encoding_info(datname text)
Description: Calls encode_feature_perf_hist to save the encoded data persistently.
Parameter: database name
Return type: int
db4ai_predict_by_bool (text, VARIADIC "any")
Description: Obtains a model whose return value is of the Boolean type for model inference. This function is an internal function. You are advised to use the PREDICT BY syntax for inference.
Parameter: model name and input column name of the inference task
Return type: Boolean
db4ai_predict_by_float4(text, VARIADIC "any")
Description: Obtains a model whose return value is of the float4 type for model inference. This function is an internal function. You are advised to use the PREDICT BY syntax for inference.
Parameter: model name and input column name of the inference task
Return type: float
db4ai_predict_by_float8(text, VARIADIC "any")
Description: Obtains a model whose return value is of the float8 type for model inference. This function is an internal function. You are advised to use the PREDICT BY syntax for inference.
Parameter: model name and input column name of the inference task
Return type: float
db4ai_predict_by_int32(text, VARIADIC "any")
Description: Obtains a model whose return value is of the int32 type for model inference. This function is an internal function. You are advised to use the PREDICT BY syntax for inference.
Parameter: model name and input column name of the inference task
Return type: int
db4ai_predict_by_int64(text, VARIADIC "any")
Description: Obtains a model whose return value is of the int64 type for model inference. This function is an internal function. You are advised to use the PREDICT BY syntax for inference.
Parameter: model name and input column name of the inference task
Return type: int
db4ai_predict_by_numeric(text, VARIADIC "any")
Description: Obtains a model whose return value is of the numeric type for model inference. This function is an internal function. You are advised to use the PREDICT BY syntax for inference.
Parameter: model name and input column name of the inference task
Return type: numeric
db4ai_predict_by_text(text, VARIADIC "any")
Description: Obtains a model whose return value is of the character type for model inference. This function is an internal function. You are advised to use the PREDICT BY syntax for inference.
Parameter: model name and input column name of the inference task
Return type: text
db4ai_predict_by_float8_array(text, VARIADIC "any")
Description: Obtains a model whose return value is of the character type for model inference. This function is an internal function. You are advised to use the PREDICT BY syntax for inference.
Parameter: model name and input column name of the inference task
Return type: text
gs_explain_model(text)
Description: Obtains the model whose return value is of the character type for text-based model parsing.
Parameter: model name
Return type: text
Example:
-- Create a data table. gaussdb=# CREATE TABLE houses ( id INTEGER, tax INTEGER, bedroom INTEGER, bath DOUBLE PRECISION, price INTEGER, size INTEGER, lot INTEGER, mark text ); -- Insert training data. gaussdb=# INSERT INTO houses(id, tax, bedroom, bath, price, size, lot, mark) VALUES (1,590,2,1,50000,770,22100,'a+'), (2,1050,3,2,85000,1410,12000,'a+'), (3,20,2,1,22500,1060,3500,'a-'), (4,870,2,2,90000,1300,17500,'a+'), (5,1320,3,2,133000,1500,30000,'a+'), (6,1350,2,1,90500,850,25700,'a-'), (7,2790,3,2.5,260000,2130,25000,'a+'), (8,680,2,1,142500,1170,22000,'a-'), (9,1840,3,2,160000,1500,19000,'a+'), (10,3680,4,2,240000,2790,20000,'a-'), (11,1660,3,1,87000,1030,17500,'a+'), (12,1620,3,2,118500,1250,20000,'a-'), (13,3100,3,2,140000,1760,38000,'a+'), (14,2090,2,3,148000,1550,14000,'a-'), (15,650,3,1.5,65000,1450,12000,'a-'); -- Train the model. gaussdb=# CREATE MODEL gs_explain USING logistic_regression FEATURES size, lot TARGET mark FROM HOUSES WITH learning_rate=0.88, max_iterations=default; -- Textualize the model. gaussdb=# select *from gs_explain_model('price_model'); gs_explain_model ---------------------------------------------------------------- Name: price_model + Algorithm: logistic_regression + Query: CREATE MODEL price_model USING logistic_regression + FEATURES size, lot + TARGET mark + FROM HOUSES + WITH learning_rate=0.88, max_iterations=default; + Return type: Text + Pre-processing time: 0.000000 + Execution time: 0.000262 + Processed tuples: 15 + Discarded tuples: 0 + batch_size: 1000 + decay: 0.9500000000 + learning_rate: 0.8800000000 + max_iterations: 100 + max_seconds: 0 + optimizer: gd + tolerance: 0.0005000000 + seed: 1719571149 + verbose: false + accuracy: 0.5333333611 + f1: 0.5333333611 + precision: 0.5000000000 + recall: 0.5714285970 + loss: -nan + weights: {1636.68648564241,-119.516647562169,1.12196231032849}+ categories: {a+,a-} + (1 row)
gs_ai_stats_explain(text, text[])
Description: Prints the intelligent statistics of multiple columns in the corresponding table and columns.
Parameters: table name and column name collection.
Return type: text
gs_acm_analyze_workload_manual()
Description: Manually trains the cardinality estimation model based on the operator feedback data in the current database.
Parameter: none
Return type: text
gs_stat_get_acm_feedback_operator_info()
Description: Displays all operator feedback data collected in the global memory.
Parameter: none
Return type: record
gs_stat_get_sql_feedback_info()
Description: Displays the number of times that SQL statements are used and the execution time when SQL statements are used and not used within the effective range of the feedback cardinality function in the global memory.
Parameter: none
Return type: record
gs_costmodel_calibration_manual()
Description: Manually triggers operator time collection and corrects cost model parameters.
Parameter: none
Return type: text
gs_show_aplan(cstring)
Description: Views the multi-plan cache of queries cached in the current session.
Parameter: prepare name is used to search for the corresponding cache plan.
Return type: text
ai_watchdog_detection_warnings()
Description: Obtains the risk alarm information of the AI watchdog. The SYSADMIN or MONADMIN access permission is required.
Parameter: none
Return type: record
ai_watchdog_monitor_status(int)
Description: Obtains the monitoring information of the AI watchdog. The SYSADMIN or MONADMIN access permission is required.
Parameter: Returns the upper limit of the length of the monitored sequence.
Return type: record
ai_watchdog_parameters()
Description: Obtains the internal parameters or status information of the AI watchdog. The SYSADMIN or MONADMIN access permission is required.
Parameter: none
Return type: record
Feedback
Was this page helpful?
Provide feedbackThank you very much for your feedback. We will continue working to improve the documentation.See the reply and handling status in My Cloud VOC.
For any further questions, feel free to contact us through the chatbot.
Chatbot
